ACI STRUCTURAL JOURNAL

TECHNICAL PAPER

Title No. 112-S49

Design versus Assessment of Concrete Structures Using

Stress Fields and Strut-and-Tie Models
by Aurelio Muttoni, Miguel Fernndez Ruiz, and Filip Niketic
Stress fields and strut-and-tie models are widely used for design
and assessment of structural concrete members. Although they
are often used in the same manner for both purposes, developing
suitable stress fields and strut-and-tie models for the design of a
new structure or for assessment of the strength of an existing one
should not necessarily be performed following the same approach.
For design, simple load-carrying models in equilibrium with the
external actions can be considered. From the various possibilities,
those leading to better behavior at serviceability limit state and
to simple reinforcement layouts should be selected (or a combination of them). For the assessment of existing structures, however,
avoiding unnecessary strengthening (or minimizing it) should be
the objective. Thus, simple stress fields or strut-and-tie models
are to be iteratively refined whenever the calculated strength of
the member is insufficient with respect to the design actions. This
can be done by accounting for kinematic considerations to calculate the higher possible strength of the member accounting for its
actual geometry and available reinforcement (allowing to calculate the exact solution according to limit analysis). In this paper,
the differences between the two approaches for design and assessment are clarified and explained on the basis of some examples.
A number of strategies are comprehensibly presented to obtain
suitable stress fields and strut-and-tie models in both cases. The
results of exact solutions according to limit analysis (developed
using elastic-plastic stress fields) are finally compared to 150 tests
of the literature showing the consistency and generality of the
presentedapproaches.
Keywords: assessment; design; limit analysis; stress fields; strut-andtie models.

INTRODUCTION
Currently, most codes for structural concrete are oriented
to design of new structures. Yet assessing the strength of
existing structures (and their retrofitting, if necessary) is
becoming more and more a relevant task for structural
engineers. This is influencing codes of practice, which start
incorporating new concepts related to life cycle design1 or
evaluation and assessment.2,3 Grounding codes of practice
on physical models reveals to be an efficient manner to cover
both aspects, as design can be performed by assuming safe
and simple hypotheses, which can be refined if necessary for
an assessment.4
Although the same physical models can be used for design
and for assessment, the use that is made of these models
can be very different depending on what type of analysis
is performed. The design of a new structure should cover
its ability to carry the design actions (safe solution at ultimate limit state [ULS]) as well as its suitable behavior
under service loads (satisfactory behavior for cracking or
ACI Structural Journal/September-October 2015

deflections for instance at serviceability limit state [SLS]).

For assessment of an existing structure, however, it is only
normally sufficient to investigate on its ultimate strength
because the SLS behavior can be usually checked in-place
by inspections and accounting for the experience of end
users (provided that the actions remain unchanged).5 In
these cases, if a conventional analysis shows an insufficient
level of safety, the estimate of the strength of the structure
needs to be more refined (accounting for other potential
load-carrying actions) because this might avoid unnecessary
retrofitting or minimizes it.
Another significant difference between a design and an
assessment refers to the concept of most suitable or optimum
solution. To achieve an optimum design, it is normally cost
that is to be optimized. In common practice, the best solution
is the one that respects ULS and SLS criteria and that allows
minimizing costs (referring to amount of materials but also
time for design and construction). In this sense, selecting
simple solutions that require a slightly higher amount of reinforcement but that are simpler to design, reinforce, control,
and cast are usually preferred. When performing an assessment of a critical structure, cost should also be optimized.
Normally, this is done if any strengthening is avoided or at
least minimized. In some cases, the requirements at ULS can
be ensured by means of simple analyses. In these cases, there
is no need to perform more refined investigations.4 However,
whenever an insufficient level of safety is found with simple
analyses, the use of more refined or advanced analysis
methods is thus plenty justified as they allow considering for
all potential load-carrying actions and provide an accurate
estimate of the actual strength of the member.
With reference to structural concrete, the use of stress
fields and strut-and-tie models provide a physically consistent frame to design new structures and to assess their
strength. They are both based on the lower-bound theorem
of limit analysis. Limit analysis can be considered as a
method to calculate the failure load of a member considering a perfectly plastic behavior of materials.6 According to
Drucker,7 the lower-bound theorem of limit analysis can be
written for reinforced concrete members as follows: If an
equilibrium distribution of stress can be found in the concrete
and the steel which is nowhere tensile in the concrete and is
ACI Structural Journal, V. 112, No. 5, September-October 2015.
MS No. S-2014-178.R2, doi: 10.14359/51687710, received October 14, 2014, and
reviewed under Institute publication policies. Copyright 2015, American Concrete
Institute. All rights reserved, including the making of copies unless permission is
obtained from the copyright proprietors. Pertinent discussion including authors
closure, if any, will be published ten months from this journals date if the discussion
is received within four months of the papers print publication.

605

Fig. 2Investigation of beam under point load according

to Drucker7: (a) stress field; and (b) corresponding
failuremechanism.
Fig. 1Stress fields and strut-and-tie models: (a) stress field
for beam near support region; and (b) corresponding strutand-tie model.
everywhere at or below yield, the structure will not collapse
or will just be at the point of collapse. It can be noted that in
this formulation, concrete crushing is intended as yielding.
This means that models for structural members according
to strut-and-tie and stress fields methods, by definition,
might underestimate the strength of the member (provided
sufficient ductility of the member). The stress field method8
allows determining a licit state of stresses in a structural
concrete member in equilibrium with a given set of actions;
refer to Fig. 1(a). On that basis, the stresses developing in the
concrete can be calculated (Fig. 1(a)) as well as the amount
and location of the tensile reinforcement. The resultants of
the stress field can further be used to develop a strut-and-tie
model, providing a view of how the forces are carried within
the member (Fig. 1(b)). The angles of the struts and ties as
well as their forces can be consistently calculated on the
basis of the stress field (refer, for instance, to the angle of
the strut corresponding to the fan region, different from the
value at the region where the compression field develops at
constant angle and allowing to consider the actual support
dimensions). Because both methods are based on the lowerbound theorem of limit analysis, they can be used to obtain
safe solutions for design of new structures (equilibrium solutions respecting the yield criteria of the materials).
Limit analysis has also an upper bound theorem which
can be very useful for calculating the load-carrying capacity
in the case of assessment of existing structures. According
to Drucker7: The structure will collapse if there is any
compatible pattern of plastic deformation for which the rate
of work of the external loads exceeds the rate of internal
dissipation. When a stress field satisfies both the lowerand upper-bound solution, it is named as the exact solution
according to limit analysis (unique if, as usually assumed,
the yield surface is convex and the increment of plastic
strains is normal to the yield surfaces of the materials9). For
instance, Drucker showed that for the stress field presented
606

in Fig. 2(a), the structure can develop a licit failure mechanism (Fig. 2(b)) and is thus an exact solution (if the structure
has sufficient ductility). Such a solution can be considered
as the best potential stress field that can be developed for the
assessment of an existing structure, as it is the one providing
the largest strength of all potential lower bounds (safe solutions). Even if the strengthening cannot be avoided with
the most refined estimate of the strength (exact solution), it
can be minimized, providing significant potential savings to
the operation.
In this paper, various strategies for developing suitable
stress fields and strut-and-tie models for design and assessment of structural concrete members are presented and
discussed with reference to some examples. The concepts
are explained from a general perspective, although the application will be limited to two-dimensional cases (walls and
beams). The accuracy of limit analysis is finally compared to
available test data showing the consistency of the approach
and its generality.
RESEARCH SIGNIFICANCE
This paper presents a discussion on the conceptual differences for the approaches that should be followed when
performing the design of a new structure and the assessment of an existing structural member using stress fields and
strut-and-tie models. For structural concrete (members with
sufficient ductility), it is shown that limit analysis and its
theorems can be consistently used for both purposes, and
a number of tailored procedures (both using hand-made or
computer-based procedures) are presented and discussed.
The suitability of limit analysis is finally validated in a
systematic manner by assessing the strength of test results
on a wide number of structural members and failure modes.
STRESS FIELDS AND STRUT-AND-TIE MODELS
As previously explained, stress fields and strut-and-tie
models share a common root as they are grounded on limit
analysis. However, the methods and their application have
a number of differences, each of them presenting some
specific advantages.

ACI Structural Journal/September-October 2015

Fig. 3Behavior of materials: (a) rigid-plastic steel; (b) elastic-plastic steel; and (c) elastic-plastic cracked concrete.
Truss models, which can be considered ancestors of
strut-and-tie models, were first developed on the basis of
intuition for beams.10,11 Truss models allowed calculating
a possible equilibrium solution where the inner forces of a
truss developing inside the concrete element equilibrate the
external actions (refer, for instance, to Fig. 1(b)). Compression forces were supposed to be carried by concrete and
tension by the reinforcement (thus allowing consistent
design). Truss models were later developed and generalized.
In Stuttgart,12-14 it was shown that there was no need for the
load-carrying model to be an actual truss and that funicular
models in equilibrium with the applied actions were suitable, leading to the development of the strut-and-tie models.
These works also opened an approach where the location
of the strut and the ties was proposed to be inspired on the
elastic uncracked stress field of a member. Such an approach
was very convenient at that time because elastic uncracked
internal forces could be calculated with available procedures
(photoelasticity or linear FEM). The consistency of strutand-tie models with the lower-bound theorem of limit analysis was also investigated,14 as well as automatic procedures
for developing suitable strut-and-tie models.15
The stress fields originated on a different basis, based
directly on the theory of plasticity. This theory can be
grounded on the works of Gvozdev (published in English
in 196016), who first formulated consistently the concepts
of the yield surface, upper- and lower-bound solutions, and
the flow rule. The theory was later developed and was used
in 1961 by Drucker7 to develop a stress field for a simply
supported beam under point load (Fig. 2(a)) or distributed
loading. The stress fields were later developed particularly in
Denmark6 and Switzerland,17 leading to a consistent method
for design of concrete structures.
For their use in practice, each method has its advantages. Using them in a combined manner is probably the
most suitable approach for the designer, as both are in fact
different manners of expressing the same physical reality.
For instance, stress fields provide a detailed description
of the stress state of the member (Fig. 1(a)). They allow
determining the width of the compression fields or struts,
thus calculating the stresses developing in the concrete to
compare them to the concrete strength (Fig. 1(a)). Also,
they allow understanding and identifying the location where
smeared reinforcement is to be arranged. The development of stress fields is thus particularly useful for detailing
(required space for the struts, reinforcement bents, stresses
in the nodal regions) and to account for variable angle of
the compression fields (fan or constant-angle compression

ACI Structural Journal/September-October 2015

fields). Despite these advantages, a complete development

of the stress field, particularly in non-critical regions, may
be time-consuming and unnecessary. This is the reason why
combining stress fields with strut-and-tie models (Fig.1(b))
is usually advantageous. The strut-and-tie models can in
fact be considered as a simplification of the stress fields,
accounting only the resultants (forces) of the stress fields.
This allows a simple calculation of the internal forces and
of the required reinforcement at the ties (concentrated or
smeared), the critical concrete regions requiring, however, a
more detailed stress field analysis.
USE OF LIMIT ANALYSIS FOR DESIGN AND
ASSESSMENT OF STRUCTURAL CONCRETE
Application of limit analysis to structural concrete requires
a number of considerations. A detailed analysis of the topic
can be consulted elsewhere.6,8,18 In the following, reference
will be made to members with sufficient reinforcement to
control cracking and avoid crack localization.
With respect to reinforcing steel, it can be modeled as a
rigid-plastic (Fig. 3(a)) or elastic-plastic (Fig. 3(b)) material.
Concrete is, however, more complex to model as a perfectly
plastic material due to its brittle behavior in compression and
to the development of cracking in tension (Fig. 3(c)). Suitable approaches for limit analysis neglect its tensile strength
and consider the material as rigid-plastic or elastic-plastic in
compression with a plastic plateau strength fcp

fcp = fc fc

where fc refers to the uniaxial compressive strength of

concrete; and fc and are strength reduction factors
accounting, respectively, for the brittleness of concrete and
the influence of transverse cracking on the strength of the
compression field. The former can be evaluated as19

f
fc = c 0
fc

1/ 3

1.0

where fc0 is a reference compressive strength that can be

set to 30 MPa (4300 psi) for normal-strength concrete, as
proposed in Reference 1. With respect to the latter factor
the influence of transverse cracking on the compressive
strengthit has been thoroughly investigated by several
researchers. For instance, Vecchio and Collins20 proposed a
suitable value for coefficient as follows

1.0
1.0
0.8 + 170 1
607

Fig. 4Development of equilibrium solutions based on simple load-carrying models (values provided in Table 1): (a) investigated member; (b) analysis as dapped-end beam; (c) and (d) consideration of contribution of column and beam around
opening; (e) combined model; and (f) required reinforcement to carry loads and welded wire of minimum reinforcement.
where 1 refers to the principal transverse strain of concrete.
This strain can be calculated when the materials are assumed
to behave with an elastic phase prior to reaching their plastic
plateau.18 When the materials are assumed to behave in a
rigid-plastic manner (typically useful for hand-made calculations), the value of parameter is set depending on the
cracking state and orientation1,8 (1.0 for uncracked concrete,
0.8 for concrete cracked parallel to the loading direction, 0.6
for diagonally cracked concrete as in shear, and 0.4 for large
plastic strains in two directions as shear near plastic hinges).
The influence of other phenomena on the compressive
strength of concrete can also be accounted; for instance, the
presence of ducts21,22 or the influence of sustained loading.23
DESIGN OF NEW STRUCTURES
As previously explained, design of new elements should be
performed to obtain safe structures at ULS with satisfactory
behavior at SLS24 and with a reinforcement layout as simple
as possible for its construction and control. Developing
suitable stress fields or strut-and-tie models to that aim can
be performed following a number of strategies. Four possibilities are presented hereafter, but others (or combinations
of them) can also apply. The selection of the most suitable
approach for a given case is usually decided by the designer,
depending on his or her experience, the complexity of the
problem, and the required level of accuracy of theanalysis.
Load-carrying models inspired on experience or
on similar cases found in literature
Very often, design of a new structural member can be
performed by assimilating it to a known case. This, for
instance, is shown in Fig. 4(a) (beam with an opening). This
608

case can be designed in a simple manner by considering it as

a dapped-end member, allowing for calculation of the necessary reinforcement (Fig. 4(b) and Table 1). If necessary, the
load-carrying model can be enhanced by considering, for
instance, the contribution of the coupling beam and column
of the member (at left and below the opening), and incorporating them into the load-carrying model (Fig. 4(c) through
(e) and Table 1). This allows simple design of the main reinforcement of the element; refer to Fig. 4(f) (calculations
performed with a design yield strength of the reinforcement
equal to 435 MPa [63 ksi]). This reinforcement is usually
completed with a minimum smeared reinforcement for crack
control and to allow for spreading of concentrated loads14,25
(refer to the welded wire reinforcement in Fig. 4(f)).
Load-carrying models inspired on deviated
thrustlines
An alternative manner to the previous approach consists
on drawing the ideal thrust lines necessary to carry the forces
to the supports (Fig. 5(a) and Table 2) regardless of the actual
geometry of the member. Then, the compression struts are
to be deviated (by placing of a suitable reinforcement) so
as to remain within the available concrete (Fig. 5(b) and
Table2). This method has been acknowledged as a general
and suitable manner to develop stress fields. Its use however
requires sometimes a number of iterations to avoid four
potential problems:
Zones without reinforcement where tensile stresses are
expected. This is, for instance, the situation presented in
Fig. 6(a). In this case, the primary tension tie is located
above the opening and large flexural cracks may develop
in an unsuitable manner below it (Fig. 6(b)). Even if a
ACI Structural Journal/September-October 2015

Table 1Forces obtained using strut-and-tie

models in Fig. 4
Forces
Strut (C) or tie (S)

kN

kip

C1

946

213

C2

632

142

C3

911

205

C4

597

134

C5

1460

328

S1

691

155

S2

620

139

S3

593

133

C6

665

149

C7

638

143

C8

844

190

C9

576

129

C10

322

72

C11

207

47

C12

248

56

C13

591

133

C14

837

188

C15

606

136

C16

1460

328

S4

99

22

S5

289

65

S6

691

155

S7

311

70

S8

461

104

S9

83

19

S10

433

97

Fig. 5Load-carrying models inspired on deviated thrust

lines (values provided in Table 2): (a) funicular strut-and-tie
model; and (b) deviation of left strut to account for presence
of opening.

Table 2Forces obtained using strut-and-tie

models in Fig. 5
Forces
Strut (C) or tie (S)

kN

kip

C1

994

223

C2

1460

328

S1

691

155

C3

634

143

C4

845

190

C5

575

129

C6

644

145

C7

871

196

C8

606

136

S2

510

115

S3

510

155

ACI Structural Journal/September-October 2015

reinforcement for crack control is placed below the tie,

significant crack widths in region A of Fig. 6(b) are
required to activate the primary tie. To avoid such cases,
Muttoni et al.8 proposed to account for the kinematics of
the element to avoid regions where cracking develops in
an uncontrolled manner. In the case of Fig. 6(b), it will
consist of analyzing qualitatively the crack pattern of
the member (Fig. 6(b)) and adapting the reinforcement
layout to have a more suitable response (shifting the
tie below the opening; for example, Fig. 5(a) and (b)).
Alternatively, investigation of the tension zones of the
uncracked stress field of the member might be useful to
this respect.14
The angle between the struts and the ties should be
checked (region B of Fig. 6(b)). When no compatibility conditions are accounted (assuming, for instance,
a rigid-plastic material behavior) and if no transverse
reinforcement is available, the angle between the struts
and the ties should usually be considered larger than or
equal to 45 degrees.8 In the presence of transverse reinforcement, the angle between the struts and the ties can
be reduced, but it is suggested not to reduce the angles
below 20 to 25 degrees. This limit is grounded26 on the
fact that, otherwise, the state of strains of the member
become rather incompatible (with tension in the reinforcement and compression strains in the concrete
developing at a low angle) requiring large tensile strains
as well as crack widths to develop, which potentially
reduces the value of coefficient below the conventional thresholds (0.6 in this case). This is the reason
why, for flat angles of the compression field, analyses
based on elastic-plastic material properties accounting
for cracking lead to more accurate results.21,22 An upper
limit to this angle can be necessary in case of potential
brittle anchorage failures.
Unsuitable nodal regions should be avoided. Generally,
TTT (tension-tension-tension) nodes are unsuitable in
strut-and-tie and stress fields models.8 This is justified by
the large crack widths potentially developing in the nodal
region, leading to very low values of the strength-reduction factor . Whenever a TTT nodal region is present
in a load-carrying model, it is preferable to modify the
topology to avoid such kind of node, or prestressing of
at least one tie is required. Other nodal regions such as
609

anchorage of tension ties. For instance, Fig.6(c) presents a load-carrying model where the vertical tension
tie is anchored within the element and not at its edge.
As experimentally shown by Maxwell and Breen,28 this
may lead to development of wide cracks on top of the
anchorage region because the strut tends to occupy all
available concrete (unreinforced region of anchorage of
the vertical tie; refer to region C in Fig. 6(d)), which
cracks in an uncontrolled manner.

Fig. 6Potential problems for simple load-carrying

models: (a) strut-and-tie model with primary tension tie
above opening; (b) corresponding cracking pattern; (c)
load-carrying model with one vertical tie anchored within
concrete element; and (d) corresponding cracking pattern.28

Fig. 7Development of strut-and-tie models on basis of

elastic-uncracked stress fields: (a) elastic-uncracked stress
field of investigated element; and (b) corresponding strutand-tie model.

Fig. 8Elastic-plastic stress fields: (a) resulting EPSF for

investigated member; and (b) calculated reinforcement.

610

CTT or CCT (where C refers to compression and T to

tension) can be used, provided that the minimum angles
between the struts and the ties and the values of coefficient previously discussed are respected. CCC nodes
can be used without anyrestriction.
The strut-and-tie model has to consider that compression in concrete tends to occupy all available space
following the St-Venants principle.13 This is particularly
significant with respect to the introduction of concentrated loads.25,27 It should also be considered for the

Load-carrying models inspired on elastic

uncracked stress field of member
A classical approach to development of strut-and-tie
models was early investigated by Leonhardt and Walther12
and later refined by Schlaich et al.14 The approach considers
the elastic uncracked stress field of a member (Fig. 7(a)) and
arranges the reinforcement following the location, the direction, and the intensity of the tensile stresses (Fig. 7(b)). This
method provides, other than a safe design, generally satisfactory behavior at SLS (cracking) because the deformation
of the zones in tension controlled by a reinforcement. This
approach, however, presents a number of shortcomings:
The location and arrangement of the reinforcement is
not decided by the designer. Usually, inclined bars result
from the elastic stress field, which are very efficient for
control of cracking but might be unpractical for placing
(refer to Fig. 7(b)).
The method leads potentially to TTT nodes (for instance,
for corner frames with opening moments29) and requires
modification of the strut-and-tie model.
The amount of required reinforcement is not necessarily
the minimum required for statical reasons. This is, for
instance, the case when imposed strains are considered
as actions (imposed strains can be dissipated provided
that the member has sufficient ductility). Also, this
results from the location of the tension ties, which are
generally more efficient when placed at the outermost
fiber of the member respecting concrete cover, to maximize the lever arms rather than at the resultant of the
uncracked stress field.
Load-carrying models inspired on cracked stress
field of member
More suitable stress fields that can be used to develop
consistent load-carrying models refer to those accounting
for the cracked behavior of concrete. To that purpose,
advanced constitutive material models can be used. Alternatively, Fernndez Ruiz and Muttoni18 proposed to develop
simple elastic-plastic stress fields (EPSFs) accounting for
the cracked behavior of concrete (Fig. 3(c)). This can be seen
as a simplification of more general constitutive models and
leads to suitable results when sufficient transverse reinforcement is available to avoid crack localization.18 Figure 8(a)
depicts the corresponding EPSF for the previously investigated case of a deep beam with an opening. The EPSF indicates the location of the compression fields, fans, and forces
carried by the reinforcement. Although strut-and-tie models
(or rigid-plastic stress fields) can be developed on that
basis,18 the results of the EPSF can be directly used for the
ACI Structural Journal/September-October 2015

design and optimization of the reinforcement30 because they

are already a licit stress field. An approach to optimize the
reinforcement layout and amount has been also presented
by Fernndez Ruiz and Muttoni.18 It consists of starting
with a preliminary analysis where a minimum reinforcement for crack control is arranged in the member (Asi,min)
with the geometry and location preferred by the designer; for
instance, the orthogonal welded wire reinforcement shown
in Fig. 8(a). The reinforcement is considered to behave
elastic and the concrete is considered to behave with no
tensile strength (cracked behavior). From this preliminary
analysis, the stresses developed in the various reinforcement
bars can be calculated (si). The area of the reinforcements
(Asi) can thus be updated for the next step to respect the yield
condition of the reinforcement

Asi , j = Asi , j 1

si
Asi ,min
f yi

where fyi refers to the yield strength of the reinforcement;

and i and j refer to the iteration number (j = 2 is the first
iteration; j = 1 is the first step corresponding to Asi,min). With
the updated reinforcement amounts, the EPSF analyses can
be repeated until the solution converges to a final reinforcement layout (Fig. 8(b); calculations performed for a design
yield strength of the tension reinforcement equal to 435 MPa
[63 ksi]), where the required diameter of the bars is slightly
increased to round them to available commercial values
(which introduces a slight strength reserve). The required
number of iterations to attain the final solution is quite low,
and the method is robust for its practical application31 (there
is no need to perform further rigid-plastic analyses). In addition, because the minimum amount of reinforcement (Asi,min)
is directly considered in the first step (and is not simply added
at the end), its contribution can be considered for calculation
of the failure load and this leads to potential savings in the
total amount of reinforcement. Additionally, taking advantage of the compatibility conditions of the EPSF, the final
reinforcement layout can be also analysed at serviceability
limit state, both for deflections and cracking.
ASSESSMENT OF EXISTING STRUCTURES
Structural assessment is usually performed when the loads
acting on a structure are increased, deficiencies related to
design or construction are known, deterioration occurs, or
when current code provisions are more severe than the ones
used for the original design of the structure. Retrofitting of
existing structures is, however, expensive and complicated
and thus the assessment of existing structures has to look
for avoiding or minimizing them. Typically, this can be
performed following a levels-of-approximation approach,4
starting with simple load-carrying models (or a combination
of them) and refining them whenever insufficient strength
isobtained.
It should also be noted that, during design, even using
rational approaches, as previously presented, some elements
are not considered within the primary load-carrying models.
This is usually the case of the minimal amount of reinforcement for crack control. This additional reinforcement should
ACI Structural Journal/September-October 2015

nevertheless be considered as part of the load-carrying mechanism for an assessment as it increases the strength of the
member in a potentially non-negligible manner.32 In addition, design based on lower-bound solutions according to
limit analysis (as stress fields or strut-and-tie models) implies
that more efficient load-carrying models may develop within
the member for the available geometry andreinforcement.
In case conventional analyses do not allow ensuring sufficient strength with respect to the design actions, refined
estimates of the strength are required. In this case, the
best lower-bound solution that can be selected is the exact
solution according to limit analysis, which provides the
maximum strength of all possible lower-bounds. Different
techniques can be used to obtain the exact solution. In the
following, two approaches will be examined: the former
based on rigid-plastic stress fields and the latter based on
elastic-plastic stress fields. To that aim, the strength of the
previous element (Fig. 4(a)) will be assessed with each
approach for the reinforcement layout designed according
to the strut-and-tie inspired on simple load-carrying models
(Fig. 4(f)).
Development of exact solutions based on rigidplastic stress fields and mechanisms
The use of rigid-plastic stress fields combined with mechanisms for the search of an exact solution has been discussed
by Muttoni et al.33 It can be performed by first selecting a
licit collapse mechanism (upper-bound solution) whose free
bodies are separated by discrete cracks and concrete hinges
(refer to Fig. 9(a)). According to the upper-bound approach
of rigid-plastic limit analysis (mechanism), all reinforcement crossing the cracks reaches its yield strength so that
tie forces at ultimate limit state can be calculated easily
(Fig. 9(b), where concrete is assumed to carry no tensile
stress). The contact zones between the free bodies can be
considered compression zones where the struts or nodal
regions develop (Fig. 9(c) through (e)).
There are basically two methods to calculate the loadcarrying capacity related to a given mechanism. The first one
is based on the principle of virtual work, where the work of
external loads (sum of scalar product of external forces and
related displacements) is equal to the internal plastic dissipation (sum of all reinforcement yielding forces and concrete
forces multiplied by their elongation or shortenings). In
the second method, the equilibrium of every free body is
treatedseparately.
With respect to the second method, solving the equilibrium equations of every free body can be performed in an
easy manner by following an iterative procedure. Alternatively, the contact forces between the free bodies can be
determined without the need of performing any iterations by
solving a system of equations relating all implied variables.
Details and implementation of such procedures is thoroughly
explained in Appendix A* for the investigated wall presented
in Fig. 4(f) (refer also to Table 3 for results).
*
The Appendix is available at www.concrete.org/publications in PDF format,
appended to the online version of the published paper. It is also available in hard copy
from ACI headquarters for a fee equal to the cost of reproduction plus handling at the
time of the request.

611

Fig. 9Assessment of strength using rigid-plastic mechanisms (values provided in Table 3): (a) failure mechanism; (b) and (c)
analysis of rigid bodies of mechanism; (d) and (e) analysis of critical struts; (f) Cremona diagram and thrust line inside rigid
body; and (g) corresponding strut-and-tie model.
The load-carrying capacity QR of the selected mechanism
(Fig. 9(a)), calculated by any of the previous procedures, is
still not proven to correspond to that of the exact solution
according to limit analysis, as other potential mechanisms
(Fig. 10(a), for instance) can develop. As this procedure
is based on the upper-bound theorem of limit analysis, the
exact load-carrying capacity QR,exact could be lower than
the calculated value QR. An interesting approach to verify
if the selected mechanism corresponds to the exact solution can be performed by combining both theorems of limit
analysis. As stated by Drucker7: Agreement of upper and
lower bounds proves that the load carrying capacity is exactly
halved. This procedure can be seen as an optimization
where the criterion is related to the load-carryingcapacity.
In the current example, this can be done by investigating
the internal equilibrium of all bodies as well as the strengths
of steel and concrete elements using strut-and-tie or stress
fields models. This is, for instance, shown in Fig. 9(g) for
the reinforcement layout of Fig. 4(f) (inspired by simple
load-carrying models). A suitable strategy can consist in
developing, within the free bodies, the previously calculated internal forces acting on the edges of the free bodies.
Starting from force CAD on concrete body A (Fig. 9(b)), the
first node NA1 can be set at the intersection with the lower
reinforcement, whose force SAB3 deviates the strut. Nodal
equilibrium can be solved analytically or graphically using
force diagrams. In this case, the second approach allows
following the strut and its deviation at every reinforcement,
until the internal force CAB on the opposite side is reached.
Provided that the compression fields remain within concrete
612

(Fig. 9(g)), a licit solution will have been obtained. On the

contrary, when the calculated stress field leads to compression forces developing where no concrete is available
(Fig.10(b)), the selected mechanism does not correspond to
the exact solution and the ultimate strength is overestimated
(Qu = 3201 kN [720 kip] in Fig. 10(b)). In this case, other
mechanisms have to be investigated.
The obtained result (Fig. 9(g)) for the reinforcement layout
of Fig. 4(f) leads to a failure load of 2408 kN (540kip). This
result is above the original design load of 2000 kN (450kip)
and shows a strength reserve of 20%. This is due to the
fact that: 1) the selected strut-and-tie model for design of
the main reinforcement is a lower bound; 2) the necessary
reinforcement was rounded (increased) to the next available commercial diameter; and 3) the minimum reinforcement amount was not considered to contribute to its strength
(contrary to the EPSF approach for design shown in Fig. 8).
Development of exact solutions based on elasticplastic stress fields
Despite the generality of the approach explained in the
previous section, obtaining a solution following a rigidplastic approach might be time-consuming in some cases.
A suitable alternative to overcoming this shortcoming is
the development of elastic-plastic stress fields that can be
used to obtain exact solutions according to limit analysis in
an automated and time-efficient manner. This type of stress
field18 considers an elastic behavior for concrete and steel
(following their elastic stiffness) until the plastic plateau
is reached. No tensile stresses are also considered for
ACI Structural Journal/September-October 2015

Table 3Forces acting in free bodies in Fig. 9

Forces

Free
body

Force
name

First iteration
(equilibrium not fulfilled)

Final iteration
(equilibrium fulfilled)

kN

kip

kN

kip

SAB1

273

61

273

61

SAB2

437

98

437

98

SAB3

273

61

273

61

SAB4

481

108

481

108

CAB,X

754

169

897

201

CAB,Y

71

16

28

CAD,X

143

32

CAD,Y

639

143

738

165

SAB1

273

61

273

61

SAB2

437

98

437

98

SAB3

273

61

273

61

SAB4

481

108

481

108

SBC1

295

66

295

66

SBC2

295

66

295

66

CAB,X

754

169

897

201

CAB,Y

71

16

28

CBC,X

590

132

447

100

CBC,Y

122

27

R2U,X

R2U,Y

1150

259

1548

348

QU,X

QU,Y

1789

402

2408

540

SCD1

295

66

295

66

SCD2

295

66

295

66

SBC1

295

66

295

66

SBC2

295

66

295

66

CBC,X

590

132

447

100

CBC,Y

122

27

CCD,X

590

132

447

100

CCD,Y

122

27

SCD1

295

66

295

66

SCD2

295

66

295

66

CAD,X

143

32

CAD,Y

639

144

738

165

CCD,X

590

132

447

100

CCD,Y

122

27

R1U,X

R1U,Y

639

144

860

192

concrete. This type of stress field, because it accounts for

the conditions of a lower-bound solution (equilibrium and
yield conditions) as well as the compatibility of displacements (stress field calculated on the basis of a displacement
ACI Structural Journal/September-October 2015

Fig. 10Analysis of other potential failure mechanisms:

(a) assumed cracks and kinematics; and (b) corresponding
(illicit) strut-and-tie model.
field), allows for obtaining exact solutions in an automated
manner. This is, for instance, shown in Fig. 11, where the
elastic-plastic stress field (Fig. 11(a)) and its corresponding
displacement field (Fig. 11(b)) are presented for the investigated member for the reinforcement layout of Fig. 4(f). The
failure load results 2680 kN (603 kip). The difference with
respect to the mechanism calculated using the rigid-plastic
approach can mostly be explained by the vertical reinforcement of the column at the left-hand side of the opening
(which was neglected for the rigid-plastic analysis) and by
the role of the compression reinforcement (also neglected
for the rigid-plastic analysis shown previously). It can also
be noted that the computed failure mechanism (Fig. 11(b))
nicely agrees with the selected mechanism at failure for the
rigid-plasticanalysis.
VALIDATION
Many works can be found on the applicability of limit
analysis to reinforced concrete showing its generality and
consistency (refer, for instance, to Reference 6). This is verified provided that no strain localization happens (minimum
amount of reinforcement required) and that the material and
structural member have a ductile behavior. In other cases,
theories accounting for crack localization34 or crushing
of concrete prior to yielding of the reinforcement6 are to
beused.
To investigate on the accuracy of EPSF (as a predictor of
exact solutions according to limit analysis), Fig. 12 presents
a comparison of the results of this tool against 150 available
test data (prestressed and reinforced beams, corner frames,
and walls with and without openings). The assessment of
the strength of the tests is performed by developing an
EPSF for each element18 (the behavior of steel is assumed
to be elastic-plastic with no strain hardening). The results of
EPSF show consistent results for all investigated cases. The
average ratio of the actual failure load to predicted is 1.05
and presents a fairly low value of the coefficient of variation
(10%, considering all the tests). Details for each series on the
specimens, failure mode, and predicted failure loads can be
found in Table B1 (refer to Appendix B) of this paper. The
results, shown in Fig. 12, show no trend for the most significant mechanical parameters (compressive strength, level of
axial load of beams, and longitudinal and transversal reinforcement ratios) and validate the generality and accuracy
of the approach.

613

Fig. 11Assessment using EPSF: (a) resulting stress field;

and (b) calculated displacement field at failure (mechanism).

Fig. 12Assessment of test results using EPSF: comparison

of actual strength and predictions as a function of compressive strength (fc ), axial force level (P/A) longitudinal reinforcement ratio () and transverse reinforcement ratio (w)
(tests detailed in Table B1, average of Qtest/QEPSF = 1.05,
CoV = 10%). (Note: 1 MPa = 145 psi.)
CONCLUSIONS
This paper discusses suitable approaches for the design
and assessment of structural concrete members by using
stress fields and strut-and-tie models. Its most significant
conclusions are summarized as follows:
1. Design and assessment of structural concrete can be
performed on the basis of limit analysis provided that its
hypotheses (mostly avoidance of strain localization which
leads to brittle behavior) are respected. However, design and
assessment should not necessarily be performed following
the same strategies. For design, simple load-carrying models
in equilibrium with the external actions and respecting the
yield conditions of the materials are usually suitable. From
the various possibilities, those leading to a better behavior at
serviceability limit state and corresponding to simple reinforcement layouts are the best choices. For assessment of
existing critical structures, the estimate of the strength has to
be refined. To that aim, developing exact solutions according

614

to limit analysis is a suitable approach (they provide the

highest strength of all potential safe estimates). This allows
avoiding unnecessary strengthening of existing structures
or minimizing it. Serviceability issues in these cases can be
checked directly on the existing structure in case the actions
remain unchanged.
2. Stress fields and strut-and-tie models are both based
on the lower-bound theorem of limit analysis and can be
efficiently used for design of new structures. Each method
presents a number of advantages. Stress fields are most
suited for detailing, checking of concrete behavior, and
providing a better estimate of the location of the struts. Strutand-tie models are simpler for developing calculation of the
requiredreinforcement.
3. Exact solutions can be developed manually in an
iterative manner by considering a failure mechanism and
by calculating the corresponding rigid-plastic stress field
inside. In case the compression field cannot develop within
the available geometry, the failure mechanism has to be
adaptedaccordingly.
4. Automated procedures based on EPSFs are suitable for
both design of new structures and for assessment of existing
structures. EPSFs satisfy equilibrium and yield conditions
(lower-bound solution), but also compatibility conditions (upper-bound solution, as at ultimate, a mechanism
develops). As a consequence, they can be used to obtain
exact solutions according to limit analysis in an automated
manner. This tool has the further advantage of accounting for
all potential load-carrying actions, particularly of minimum
reinforcement, and to allow refined estimates of the strength
parameters of concrete (as the reduction on the concrete
strength due to transverse cracking).
5. The EPSF method allows calculating directly the
load-carrying capacity of an existing structure and thus
assessing its strength. The EPSF method can also be used
for designing new structures according to a simple iterative
procedure described within this paper.
6. A systematic comparison of exact solutions according to
limit analysis to available test data (150 tests) shows consistent results for a wide number of geometries (such as dapped
end beams, deep beams, prestressed girders, corner frames,
and walls) and failure modes (including shear, bending, and
detailing), providing accurate estimates of their strength
with a low scatter of the predictions, and proving the consistency and generality of the approach.
AUTHOR BIOS

ACI member Aurelio Muttoni is a Professor and Head of the Structural

Concrete Laboratory at Ecole Polytechnique Fdrale de Lausanne (EPFL),
Lausanne, Switzerland. Muttoni received the ACI Chester Paul Siess Award
for Excellence in Structural Research in 2010 and was a co-recipient of the
ACI Wason Medal for Most Meritorious Paper in 2014. His research interests include the theoretical bases of the design of reinforced concrete structures, shear and punching shear, fiber-reinforced high-strength concrete,
soil-structure interaction, and the conceptual design of bridges.
Miguel Fernndez Ruiz is a Lecturer and Senior Scientist at EPFL. His
research interests include the serviceability behavior of structures, bond,
shear and punching shear, and the modeling of structural concrete using
stress fields.
Filip Niketi is a Research Assistant and PhD Candidate at EPFL. He was
a co-recipient of the ACI Wason Medal for Most Meritorious Paper in 2014.

ACI Structural Journal/September-October 2015

His research interests include the applicability of stress fields for design and
assessment of structural concrete.

REFERENCES

1. fib Model Code for Concrete Structures 2010, Fdration Internationale du Bton, Lausanne, Switzerland, 2013, 434 pp.
2. ACI Committee 437, Strength Evaluation of Existing Concrete
Buildings (ACI 437R-03), American Concrete Institute, Farmington Hills,
MI, 2003, 28 pp.
3. ACI Committee 364, Guide for Evaluation of Concrete Structures
before Rehabilitation (ACI 364.1R-07), American Concrete Institute,
Farmington Hills, MI, 2007, 18 pp.
4. Muttoni, A., and Fernndez Ruiz, M., Levels-of-Approximation
Approach in Codes of Practice, Structural Engineering International,
V.22, No. 2, 2012, pp. 190-194. doi: 10.2749/101686612X13291382990688
5. SIA 269:2011, Existing Structures Bases, Swiss Society of Engineers and Architects, Zurich, Switzerland, 2011, 28 pp.
6. Nielsen, M. P., and Hoang, L. C., Limit Analysis and Concrete Plasticity, third edition, CRC Press, Boca Raton, FL, 2011, 796 pp.
7. Drucker, D. C., On Structural Concrete and the Theorems of Limit
Analysis, International Association for Bridge and Structural Engineering,
Zrich, Switzerland, 1961, 36 pp.
8. Muttoni, A.; Schwartz, J.; and Thrlimann, B., Design of Concrete
Structures with Stress Fields, Birkhuser, Basel-Boston-Berlin, Switzerland, 1997, 143 pp.
9. Chen, W.-F., and Han, D.-J., Plasticity for Structural Engineers,
J.Ross Publishing, Fort Lauderdale, FL, 2007, 255 pp.
10. Ritter, W., Die Bauweise Hennebique, Schweizerische Bauzeitung,
V. 33, No. 7, 1899, pp. 41-61.
11. Mrsch, E., Der Eisenbetonbau, seine Theorie und Anwendung, third
edition, Verlag von Konrad Wittwer, Hamburg, Germany, 1908, 376 pp.
12. Leonhardt, F., and Walther, R., Wandartige Trger, Deutscher Ausschuss fr Stahlbeton, Berlin, Germany, 1966, 159 pp.
13. Schlaich, J., and Weischede, D., Ein praktisches Verfahren zum
methodischen Bemessen und Konstruieren im Stahlbetonbau, Bulletin
dInformation No. 150, Comit Euro-International du Bton, Lausanne,
Switzerland, 1982, 163 pp.
14. Schlaich, J.; Schfer, K.; and Jennewein, M., Toward a Consistent Design of Structural Concrete, PCI Journal, V. 32, No. 3, 1987,
pp.74-150. doi: 10.15554/pcij.05011987.74.150
15. Biondini, F.; Bontempi, F.; and Malerba, P. G., Stress Path Adapting
Strut-and-Tie Models in Cracked and Uncracked R.C. Elements, Structural Engineering & Mechanics, V. 12, No. 6, 2001, pp. 685-698. doi:
10.12989/sem.2001.12.6.685
16. Gvozdev, A. A., The Determination of the Value of the Collapse
Load for Statically Indeterminate Systems Undergoing Plastic Deformation, International Journal of Mechanical Sciences, V. 1, No. 4, 1960,
pp.322-335. doi: 10.1016/0020-7403(60)90051-5
17. Thrlimann, B.; Marti, P.; Pralong, J.; Ritz, P.; and Zimmerli, B.,
Anwendung der Plastizittstheorie auf Stahlbeton, Institut fr Baustatik
und Konstruction, ETH Zrich, 1983, 252 pp.
18. Fernndez Ruiz, M., and Muttoni, A., On Development of Suitable
Stress Fields for Structural Concrete, ACI Structural Journal, V. 104,
No.4, July-Aug. 2007, pp. 495-502.

ACI Structural Journal/September-October 2015

19. Muttoni, A., Die Anwendbarkeit der Plastizittstheorie in der

Bemessung von Stahlbeton, Birkhuser Verlag, Institut fr Baustatik und
Konstruktion, ETH Zrich, Basel, Switzerland, 1990, 176 pp.
20. Vecchio, F. J., and Collins, M. P., The Modified Compression-Field
theory for Reinforced Concrete Elements Subjected to Shear, ACI Journal
Proceedings, V. 83, No. 2, Mar.-Apr. 1986, pp. 219-231.
21. Fernndez Ruiz, M., and Muttoni, A., Shear Strength of ThinWebbed Post-Tensioned Beams, ACI Structural Journal, V. 105, No. 3,
May-June 2008, pp. 308-317.
22. Rupf, M.; Fernndez Ruiz, M.; and Muttoni, A., Post-Tensioned
Girders with Low Amounts of Shear Reinforcement: Shear Strength
and Influence of Flanges, Engineering Structures, V. 56, Nov, 2013,
pp.357-371. doi: 10.1016/j.engstruct.2013.05.024
23. Fernndez Ruiz, M.; Muttoni, A.; and Gambarova, P., Relationship between Nonlinear Creep and Cracking of Concrete under Uniaxial
Compression, Journal of Advanced Concrete Technology, V. 5, No. 3,
2007, pp. 383-393. doi: 10.3151/jact.5.383
24. Lourenco, M. S., and Almeida, J. F., Adaptive Stress Field Models:
Formulation and Validation, ACI Structural Journal, V. 110, No. 1,
Jan.-Feb. 2013, pp. 71-81.
25. Bayrak, O., and Brown, M., Minimum Transverse Reinforcement
for Bottle-Shaped Struts, ACI Structural Journal, V. 103, No. 6, Nov.-Dec.
2006, pp. 813-822.
26. Grob, J., and Thrlimann, B., Ultimate Strength and Design of Reinforced Concrete Beams under Bending and Shear, IABSE, V. 36, 1976,
pp.105-120.
27. Brown, M. D.; Sankovich, C. L.; Bayrak, O.; and Jirsa, J. O.,
Behavior and Efficiency of Bottle-Shaped Struts, ACI Structural Journal,
V. 103, No. 3, May-June 2006, pp. 348-354.
28. Maxwell, B. S., and Breen, J. E., Experimental Evaluation of Strutand-Tie Model Applied to Deep Beam with Opening, ACI Structural
Journal, V. 97, No. 1, Jan.-Feb. 2000, pp. 142-149.
29. Campana, S.; Fernndez Ruiz, M.; and Muttoni, A., Behaviour
of Nodal Regions of Reinforced Concrete Frames Subjected to Opening
Moments and Proposals for their Reinforcement, Engineering Structures,
V. 51, 2013, pp. 200-210. doi: 10.1016/j.engstruct.2013.01.029
30. Najafian, H. A., and Vollum, R. L., Optimising Reinforcement
Design in D Regions Using Non-Linear Finite-Element Analysis, Magazine of Concrete Research, V. 65, No. 4, 2013, pp. 234-247. doi: 10.1680/
macr.12.00063
31. Kostic, N., Topologie des Champs de Contraintes pour le Dimensionnement des Structures en Bton Arm, PhD thesis, Ecole Polytechnique Fdrale de Lausanne, Lausanne, Switzerland, 2009, 235 pp.
32. Kuchma, D.; Yindeesuk, S.; Nagle, T.; Hart, J.; and Lee, H. H.,
Experimental Validation of Strut-and-Tie Method for Complex Regions,
ACI Structural Journal, V. 105, No. 5, Sept.-Oct. 2008, pp. 578-589.
33. Muttoni, A.; Kostic, N.; and Fernndez Ruiz, M., Discussion of
paper Factors Affecting Strength of Elements Designed Using Strutand-Tie Models, ACI Structural Journal, V. 105, No. 2, Mar.-Apr. 2008,
pp. 233-237.
34. Muttoni, A., and Fernndez Ruiz, M., Shear Strength of Members
without Transverse Reinforcement as Function of Critical Shear Crack
Width, ACI Structural Journal, V. 105, No. 2, Mar.-Apr. 2008, pp. 163-172.

615

NOTES:

616

ACI Structural Journal/September-October 2015